JP2517398B2 - Frequency spectrum analysis method - Google Patents

Description

The present invention relates to a method for analyzing a frequency spectrum of a digital spectrum analyzer (digital frequency analyzer), and more particularly to continuously generated frequency domain information (frequency range information). ), How to recognize, remember and analyze.

[Prior Art and Problems to be Solved by the Invention] A digital spectrum analyzer can acquire data faster than its data processing speed. Further, as the data processing function of the spectrum analyzer is improved, the processing speed becomes faster, and the display speed cannot follow the processing speed. This display speed problem causes the data display to be discontinuous. That is, while one set of data is being displayed, even if the next set of data is fetched and processed, the display mechanism is not ready to display the next set of data, and this data cannot be displayed. Will be lost to. Therefore, the data displayed adjacently or continuously is actually scattered data with time gaps.

Typically, spectrum analyzers use triggers to start or stop data acquisition. However, these triggers are triggers based on time domain information (time domain information) regarding the input signal to be analyzed, or triggers obtained from an external source, and not triggers obtained from frequency domain information. Some spectrum analyzers have a means of recognizing spectrum events for the purpose of pass / fail judgment, but there was no spectrum analyzer that recognizes spectrum events in order to trigger data acquisition or data storage. .

Therefore, the events defined in the frequency domain,
That is, a method of recognizing a spectrum event is desired. That is, in this method, in response to recognition of a spectrum event or a command from an operator, for analysis,
It is desirable to be able to accumulate and display all data for the specified specific period. This allows the operator of the up spectrum analyzer to observe and analyze multiple frequency spectra that represent a continuous history of signal changes in the time domain for a period of interest.

Therefore, an object of the present invention is to enable a spectrum analyzer capable of continuously generating an input frequency spectrum at a speed higher than the speed at which it can be displayed in real time, to display a time-continuous frequency spectrum on one screen for analysis. To provide a method for analyzing a frequency spectrum.

[Means and Actions for Solving the Problems] A frequency spectrum representing a temporally continuous recording of a signal to be analyzed is accumulated in a memory in a target specific period, and this accumulation is continued until a predetermined amount of frequency spectrum is accumulated. To be Next, the stored frequency spectrum is read from the memory and displayed on a device capable of displaying a plurality of frequency spectra along the time axis. Markers are provided to make delta time measurements between different spectra on the screen. The continuous storage of the circular memory can be stopped by the detection of a predetermined spectrum event. Therefore, it is possible to analyze all of the spectra that continuously represent the frequency domain change of the electric signal that changes with time as one group during the target specific period.

Therefore, the present invention is a method for analyzing a frequency spectrum of a spectrum analyzer capable of continuously generating a frequency spectrum of a time-varying input signal at a speed higher than a speed capable of displaying in real time. Fourier transform processing is performed to successively generate a frequency spectrum frame composed of amplitude values of each spectrum of a predetermined number of frequencies over a selected frequency span, and the generated frequency spectrum frames are sequentially stored in a memory means. Then, a comparison amplitude value is set for the predetermined number of frequencies of the specific frequency span, and while the frequency spectrum is stored, the amplitude value for the predetermined number of frequencies is compared with the comparison amplitude value, and When it is detected that the amplitude value exceeds the comparison amplitude value, a trigger is generated and the trigger signal The stop of the storage operation of the frequency spectrum frame in the memory means is controlled based on the occurrence of the above, and each time-sequential frequency spectrum frame stored in the memory means is color-coded into an amplitude value on the frequency axis. And a method for analyzing a frequency spectrum, wherein the frequency spectrum is displayed on the display means in a frequency vs. time display format which is arranged along the time axis.

[Example] A spectrum analyzer is an electronic device that displays an electric signal in the frequency domain. In a digital spectrum analyzer, a series of amplitude measurements of the signal under analysis are taken over time and the measurements are converted into digital words. The digital word string represents changes in signal amplitude in the time domain. Fourier transform of this time domain information yields a frequency domain representation from the same information. The frequency domain representation of the electrical signal under analysis is displayed as a graph of signal power frequency. This display can be used to analyze the spectral content of the signal.

FIG. 1 shows a digital spectrum embodying the present invention.
Shows the analyzer. The time-varying electronic signal to be analyzed is input to the analog input section (10) of the spectrum analyzer. The analog input section (10) attenuates and amplifies according to the level of the input signal, and further gates the input signal,
Or it has the ability to respond to external (time domain) triggers. The analog input section (10) also includes a low-pass filter for antialiasing, which is necessary for the sampling function. An analog-to-digital converter (hereinafter referred to as ADC) (20) converts the input signal in good condition into a series of digital words. This input signal is 25.6 megasamples / sample to obtain an effective input signal band of 10MHz.
Sampled at a rate of seconds.

Digital down converter (30) is 500Hz and
The input band of 10MHz is lowered by a selectable frequency between 9.9995MHz, low pass filter processing is performed on the resultant composite signal, and the appropriately thinned filter output is processed by a fast Fourier transform (FFT) processor. Supply to a uniform filter bank (40) including.

To transform time domain information into a frequency domain representation, a digital spectrum analyzer performs a Fourier transform with some modification of what is known as the discrete Fourier transform (DFT). Usually this DFT is F
It is done using one of the algorithms known as FT. For FFT and its variants, numerous patent specifications,
It is described in textbooks and papers in the literature. But FF
All methods of performing T require a certain minimum amount of time and some amount of computational information. In the device described here, the uniform filter bank (40) has this processing function.

The output of the uniform filter bank (40) is a collection of 1024 pairs of digital words that represent the frequency distribution of the input signal in a complex number over 1024 frequency bins of a spectrum frame. (Here, the "bin" refers to a memory location where the amplitude value for each frequency is stored.) The spectrum data in this format is then input to the power and phase calculation circuit (50). . This circuit has a power and phase angle of 10 in the spectrum frame.
Generate 1024 power levels and 1024 phase angles representing each of the 24 bins.

The real-time interface (60) receives the above-mentioned complex signal pair and spectrum information in both power and phase formats and enables output to other devices in various formats. What is important for the present invention is that this real-time interface (60) performs event detection by comparing the power information with two comparison masks. This event detection occurs when the amplitude of one of the observed spectra is greater than the maximum limit value or less than the minimum limit value set by the operator.

The operator defines the event by various means using a front panel keyboard (110) or other terminal device connected to the CPU (100) by an appropriate communication port. FIG. 2 shows an example of the event definition mask displayed on the screen. This particular event definition mask is generated using the max hold function of the display formatter (70) and accumulates a history of maximum amplitude changes at all frequencies within a particular frequency span. Other methods of defining an event definition mask include moving one cursor to draw the mask, typing the definition with the keyboard, and adding a constant offset to the stored spectrum waveform.

The CPU (100) sends two event definition masks for the minimum value and the maximum value to the real-time interface (60) in response to the operator's event definition. When the spectrum information, especially the power information, is output from the power and phase calculation circuit (50), the real-time interface (60) compares this power information with the event definition mask, and the amplitude in any bin has the maximum value. It is judged whether it is larger or smaller than the minimum value, and if the maximum and minimum ranges are exceeded, an event detection output is generated. Although this event detection output can be used externally, it can also be used internally to stop and start the storage of spectrum information.

The display formatter (70) performs various data reduction calculations regarding the power and phase information amount from the power and phase calculation circuit (50). The display mechanism, as well as the human eye and perception, can respond to a new display every 200 μs (optimal speed that can be generated with a uniform filter bank (40)) so the operator can view the data in real time. To recognize
It is necessary to make some kind of data reduction. Therefore, the display formatter (70) performs data reduction in one of several ways that the operator can select. Frames may be discarded according to some decimation factor R. Alternatively, it is possible to generate a frame that is averaged every several frames and summarized for each average value. In addition, peaks or minimums are also selected.

All of the above techniques use a uniform filter bank (40)
It uses certain data compression techniques to allow the display and the operator to accompany the real-time spectrum calculations performed by the operator, and thus these methods allow the operator to account for all spectrum changes from frame to frame. It is not possible to examine it in detail.

According to the present invention, it is possible to investigate all the changes in the spectrum that occur during the target specific period. Rather than compressing or decimating the frequency domain data, or even displaying it at time intervals, the frequency domain data is accumulated in full to be displayed slowly later. In addition, we will define the spectrum event,
The operator can define which data in the frequency domain to measure. In addition, this frequency domain data can represent the changes that occur in the time domain without any time gaps, so that delta time measurements and detailed operator analysis can be performed on the complete history of spectral changes over the desired time period. Can be done by

Spectral information from the power and phase calculation circuit (50) when the operator selects "block mode" which allows the user to see all of a single spectral frame instead of a representative sample or compressed data All of the VME links (80) through the display formatter (70)
Sent to One spectrum frame of data containing 1024 bins has two 16-bit values for each bin. One of the two 16-bit values is power information and the other is phase information.

The VME link (80) discards some of this power and phase information and reorganizes the rest to form the VMEbus (90).
Send to The bins near the beginning and end of the 1024-bin frame contain data that is partially corrupted by the leakage and characteristics of the anti-aliasing filter in the analog input (10), so the middle of each spectrum frame. It is desirable to display only 800 bins of. Therefore, 224 bins of information, 112 bins at each end of each frame, are discarded before the data is sent to the CPU (and fast memory) (100) for final display.

The remaining 800 bins of data in this frame are reorganized by the VME link (80) before being transferred. This data is output from the power and phase calculation circuit (50) so that the power and phase angle values for a particular bin appear adjacent to each other in the data string. When the power and phase angle values are transferred from the VME link (80), they are rearranged so that all 800 power values are transferred in sequence, followed by all 800 phase angle values.

In addition to data reorganization and selection, the VMEbus (80) provides 32 bits of time stamp information for each frame of spectrum data. The time stamp information is generated from a reference oscillator that is controlled by the accuracy of the system. From the relationship between the individual spectra and their time of occurrence, calculations can later be made to measure the delta time. These times are also used to generate relative times to represent the time axis of the screen.

In block mode to process the spectrum, VME
When formatted by the link (80) and time stamped, the continuously generated spectrum data sequence is converted to V by direct memory access (DMA).
It is sent to the high-speed static memory in the CPU (and high-speed memory) (100) via the ME bus. After a sufficient number of frames have been stored in the high speed memory, the CPU (100) produces a display image for transmission to the display controller (120). As shown in FIG. 3, the display image generated by the CPU (100) consists of one or two separate display windows, labels, readers, color scales, and various other information. There is. The display controller (120) has a color display (130) for scrolling an image according to an instruction from the CPU (100).
And receives update information from the CPU (100) for controlling the image and changing the content of the image.

Even if the FFT calculation of the spectrum is done at a speed that cannot display the color spectrogram,
The continuous spectrum information is sent to a high speed memory attached to the CPU (100), which operates at a speed sufficient to store all the information. In addition, time domain data capture and FFT calculations can proceed while the output of the calculation result is sent to the circular portion of the address space of the memory. Then, due to the generation of a command from the operator or the detection of a defined spectrum event, the overwriting of the circular portion of the memory is performed by the desired delay time and the content of the circular portion of the memory displayed in one batch. You can stop after a while.

After the event is detected, by changing the number of the spectrum frame stored in the ring memory, in the display area of one continuous spectrum data,
You can control the location of spectrum events. This allows the operator of the spectrum analyzer to observe changes in the spectrum before the defined event, changes in the spectrum after the defined event, or a comparison of the two.

Techniques for locating trigger positions in prefilled circular data acquisition memory are well known, for example in the time domain of logic analyzers, but use similar techniques to define cumulative data in the frequency domain as the trigger mechanism. Is new.

In Figure 3, the color spectrogram display consists of a number of spectra arranged along the vertical time axis (300), the frequency of these spectra being the horizontal axis (31
0) Shown above. The amplitude value of the spectrum of each frequency is shown for each color. On the time axis (300), the time interval between the time stamp of the spectrum at a specific point on the spectrogram display and the time stamp of the spectrum at the lower end of this display is shown.

The two markers, the first marker and the second marker (330), are used to recognize a specific position on the spectrogram display. For each position where a marker is placed on the spectrogram display, the reader (340) shows frequency, amplitude and time values. In the area labeled DELTA in the reading section (340), the difference in frequency, amplitude and time between the positions of the two markers is shown. The operation knob for controlling the movement of the marker is provided at another place (not shown).

FIG. 4 shows the dual display function available in the spectrum analyzer described herein. With this display function,
An amplitude vs. frequency display (410) and a spectrogram display (400) aligned vertically are shown. Therefore, the two displays have equal frequency axes (420) and (430).
Are arranged along. The scale and center of this frequency axis can be commonly controlled. The vertical axis of the spectrogram display is the time axis (440), while the vertical axis of the amplitude versus frequency display is the amplitude axis (450). In this dual display, the amplitude and frequency display (410) is a detailed display of the current bottom edge of the color spectrogram display (400). The color spectrogram display shows the amplitude vs. frequency display (41
It is shown as a summary history of the contents of 0).

The user interface described above is implemented in the C programming language running on the UNIX system V operating system using the 68030 processor. However, the details of this implementation are not important to the present invention, so
Omit the explanation.

The apparatus described above provides improved functionality for recognizing, storing and analyzing successively generated frequency domain information. However, it is obvious to those skilled in the art that the present invention is not limited to the above-mentioned specific embodiments, and various modifications can be made without departing from the gist of the present invention.

EFFECTS OF THE INVENTION According to the present invention, the storage operation of the frequency spectrum frames successively generated in response to the detection of the spectrum event set with respect to the amplitude value of the spectrum in the memory means is stopped, and the memory is stopped. Each frequency spectrum frame stored in the unit is displayed in a frequency-versus-time display format in which amplitude values are color-coded on the frequency axis and arranged along the time axis, so that it is related to a specific spectrum event. It is possible to observe and analyze the continuous change of the frequency spectrum during a predetermined period.

[Brief description of drawings]

FIG. 1 is a block diagram of a digital spectrum analyzer for carrying out the present invention, and FIG. 2 is a spectrum spectrum analyzer.
Diagram showing amplitude-versus-frequency display with defined events, 3rd
The figure shows a color spectrogram display in which data is displayed according to the present invention, and FIG. 4 shows a dual display screen including both a color spectrogram display and an amplitude versus frequency display.

Claims (1)

(57) [Claims] 1. A method for analyzing a frequency spectrum of a spectrum analyzer capable of continuously generating a frequency spectrum of a time-varying input signal at a speed higher than that capable of being displayed in real time. Fourier transform processing is performed to successively generate a frequency spectrum frame composed of amplitude values of each spectrum of a predetermined number of frequencies over a selected frequency span, and the generated frequency spectrum frames are sequentially stored in a memory means. , Setting a comparison amplitude value for the predetermined number of frequencies in the specific frequency span, and comparing the amplitude value for the predetermined number of frequencies with the comparison amplitude value while storing the frequency spectrum When it is detected that the value exceeds the comparison amplitude value, a trigger is generated and the trigger signal is generated. On the basis of the occurrence, the stop of the storage operation of the frequency spectrum frame in the memory means is controlled, and each time-sequential frequency spectrum frame stored in the memory means is color-coded on the frequency axis. A method of analyzing a frequency spectrum, characterized by displaying on a display means in a frequency versus time display format arranged along a time axis.